Wall Thickness and Stone Seats: The Two Numbers That Decide Buildability

When a design comes back "not buildable", the verdict usually traces to two numbers: the wall thickness where a stone has to sit, and the depth of the seat that stone needs. Most designs that fail on the bench fail here first — long before any question of skill arises.

Why these two numbers decide everything

A stone held in a cut seat is held by the metal around it. The seat has to be deep enough for the stone to sit below the surrounding surface, and the wall around the seat has to be thick enough to be cut into without breaking through. When a design puts thin walls and deep-set stones in the same place, the geometry contradicts itself — and no amount of bench work resolves a contradiction that was drawn into the spec.

This is why a manufacturability review reads a drawing differently from a design review. The design question is "does it look right". The manufacturing question is "where does the metal run out" — and the honest answer to that question is usually two or three specific locations, not the whole piece.

The seat is a cut, not a hole

Seat cutting removes metal. That sounds obvious, but it is the step most designs forget: by the time the setter starts, the wall around every stone has already been reduced by exactly the volume the seat took out. A wall that looked adequate in the drawing is the wall after cutting that matters, and pierced designs reduce it twice — once for the piercing, once for the seat.

This is where openwork and setting collide. On the openwork flower ring, the seats sat inside the piercings that made the design work — the same features that made it beautiful were the features that blocked the tool path to the seat. The piece was resolved by dividing it so every seat sat exposed on its own section, fabricated separately and joined. That is a routing decision, and it was available because the wall-and-seat problem was read before metal was committed.

What the review actually looks at

A real review happens in the 3D model, not on the pretty render. Section views through the stone positions show the remaining wall as a number, not an impression. The model answers three questions:

• Where does the wall run out? The thinnest section adjacent to each seat, with the seat volume already subtracted.
• Can the tool reach? Whether the seat can be cut from the approach side at all, or whether the geometry blocks the graver before depth becomes the problem.
• What is left to hold the stone? The wall that remains to be raised into beads, or to carry a prong, after the cut.

A printed prototype then lets the seat be judged in the hand — which is where "the wall looks fine on screen" turns into "the wall flexes" more often than anyone expects. This is also the cheapest moment the whole project will ever offer for changing the answer.

What you decide vs what the workshop derives

There are two honest ways to spec this. Either you state a minimum wall thickness and let the workshop derive everything else around it — common when the piece has to hit a weight target or survive a known wear case. Or you specify the look and the stone sizes and let the workshop derive the structure — common when the design is the point and the structure is a means.

Both work. What does not work is silence: a drawing with stone sizes but no stated position for the metal, reviewed by nobody. The supplier who returns marked-up section views has done the work; the supplier who returns a price against the same drawing has priced an intention.

Red flags in your own drawing

Before you send a design, three self-checks catch most of the problems:

• Stone table vs available wall. A seat consumes more width than the stone's face suggests — the cut has to hold the stone below the surface, not just let it sit on top.
• Piercings adjacent to seats. Any opening near a stone position shares its wall with the seat. The wall is counted once, not twice.
• "As thin as possible" specs. Thin is a look, not a dimension. State the look; let the workshop state the minimum the look can survive.